Paperclip

Overview

Paperclip is an AI agent management platform designed to organize, orchestrate, and govern AI agents in a work environment. It reframes the way teams interact with AI: instead of prompting isolated chatbots or chaining together brittle automations, users operate a structured organization of agents — complete with an org chart, roles, goals, tasks, budgets, and agent templates. The platform is model-agnostic, meaning any agent runtime — from Claude, Codex, Cursor, and Gemini to open-source agents like OpenClaw or Pi — can be onboarded as long as it can receive a heartbeat. This creates a single operational layer where every agent reports into a shared mission.

Core to Paperclip is the concept of running AI agents like a company. Users define a business goal, “hire” agents for roles such as CEO, CTO, engineers, and marketers, and then approve the strategy before execution begins. The platform provides hierarchical org charts with reporting lines and job descriptions, ensuring every agent has clear responsibilities. Goal alignment is enforced at every level: projects trace back to the mission, tasks trace to projects, and agents receive contextual instructions that explain what to do and why. A built-in SKILL.md mechanism lets agents discover the context they need. Heartbeats keep the workforce active — agents wake on schedules, check their work, and perform delegated tasks automatically. Delegation flows both up and down the org chart, and cross-team requests are routed to the most suitable agent.

Paperclip also emphasizes cost control and accountability. Each agent is assigned a monthly budget with hard limits; once the limit is reached, the agent automatically stops, preventing runaway token spend. The platform tracks costs per agent, task, project, and goal, providing visibility into which operations burn resources. Communication happens through a structured ticket system, where every instruction, response, tool call, and decision is recorded with full tracing. An immutable append-only audit log ensures nothing can be edited or deleted. Governance features keep humans in charge: users can approve hires, override strategies, pause, or terminate any agent at any time.

As an open-source, MIT-licensed project, Paperclip is extensible and adaptable. Teams can add capabilities through drop-in extensions without touching core code, connect any LLM provider or tool stack, and deploy to any target. Whether you are a founder building an autonomous business, a team automating dev, content, marketing, QA, research, or outreach, or a developer looking for a polished agent orchestration layer, Paperclip provides the interface, controls, and trust needed to move from prompting AI to managing a team.

Core capabilities

Bring Your Own Agent

Connect any AI agent — Claude, Codex, Cursor, Gemini, OpenClaw, and more — into a single organizational structure without provider lock-in.

Org Chart Management

Define hierarchies, roles, and reporting lines so every agent has a boss, title, and job description.

Goal Alignment

Trace every task back to the company mission; agents receive context that explains what to do and why, supported by SKILL.md discovery.

Heartbeat Scheduling

Agents wake on a schedule, check their work, and act proactively; delegation flows automatically across the org chart.

Cost Control

Set monthly budgets per agent; hard limits automatically stop agents when they hit limits, preventing runaway spend.

Ticket System & Audit Log

Communicate through structured tickets with full tool-call tracing and an immutable append-only audit log.

Governance & Approval

Approve hires, override strategies, and pause or terminate agents at any time; you remain in control.

Open Source & Extensible

MIT-licensed codebase with drop-in extensions, adaptable to any LLM provider, tool stack, or deployment target.

Use cases

Orchestrating AI agents across development, content, social media, marketing, QA, research, and outreach

Running a fully autonomous business where agents hold defined roles and report to human leadership

Managing cross-functional agent teams under one org chart with shared goals

Controlling AI operating costs with per-agent budgets and hard spend limits

Auditing agent decisions through full tool-call tracing and an immutable ticket history

Building a custom, extensible agent management system on an open-source core

Transitioning from prompting individual AIs to managing a cohesive team of specialized agents